Generate a Gaussian TDMM simulation dataset

Generate Gaussian simulation data with or without a subject-specific random intercept.

Usage

data.gen.tdmm.gauss(n.subject = 200, n.time = 20, sigma2.b = 0.5, sigma2.e = 0.1, seed = NULL, random_intercept = TRUE)

Arguments

n.subject

Number of subjects. If omitted, the default is 200 with a random intercept and 300 without one. With random_intercept = FALSE, this is also the number of independent observations. An explicitly supplied value is used in either case.

n.time

Number of observations per subject when random_intercept = TRUE; unused when FALSE.

sigma2.b

Random-intercept variance when random_intercept = TRUE; unused when FALSE.

sigma2.e

Gaussian residual variance.

seed

Optional random seed.

random_intercept

A single logical value. TRUE retains the balanced repeated-measures simulation. FALSE generates one independent observation per subject, with no random intercept.

Value

A data frame with subject.ID, time, y, and x1. The random_intercept = FALSE case also has the additional covariate x2.

Details

When random_intercept = FALSE, time and an independent variable u are each Uniform(0, 1), x1 is standard normal, and x2 = sin(2 * pi * u) + e with e normal with standard deviation 0.25. The coefficient functions are sin(2 * pi * time), cos(2 * pi * time), and 1.5 * (time - 0.5) for the intercept, x1, and x2, respectively. Residual errors are independent normal with variance sigma2.e.

Examples

set.seed(1)
d <- data.gen.tdmm.gauss(n.subject = 5, n.time = 4,
  random_intercept = TRUE)
d.independent <- data.gen.tdmm.gauss(n.subject = 10,
  random_intercept = FALSE, seed = 1)